In most normal applications, the only thing that would change is what your features are. For example, if you wanted to find Waldo using the shape of his face and/or hat, you would probably just find some SIFT points (or something), and then build an eigenWaldoface, possibly using a PCA'd set of Waldo faces and hats as examples, and then SIFT the image and look for the places that are most like the eigenWaldoface.
This article is not interesting because it's an amazing new algorithm or something that solves some important world problem. It's interesting because it takes something that is not known among the general hacker population for doing this sort of thing really easily, and accomplishes it in a fairly simple way.
Ha, I wasn't being a grump...these kinds of problems are An important party of the evidence in showing thre practicality and usefulness of code. I love it.
I mostly wanted to see who else remembered that particular Waldo puzzle...it was the final one in one of the books
This article is not interesting because it's an amazing new algorithm or something that solves some important world problem. It's interesting because it takes something that is not known among the general hacker population for doing this sort of thing really easily, and accomplishes it in a fairly simple way.
Don't be a grump, this is cool. :(